Significance: Crime, substance use problems, suicide, and severe mental illness, such as schizophrenia and bipolar disorder, are associated with enormous personal distress and societal costs. Identifying the causes of these problems, especially very early in life, is a crucial research endeavor. Researchers have identified numerous putative environmental factors that are associated with increased risk for these severe problems, including maternal smoking during pregnancy, maternal stress during pregnancy, maternal age at childbearing, and advancing paternal age at childbearing. We do not know, however, whether these risk factors are casual risk factors for severe forms of psychopathology or are merely indices of family confounding. Our limited understanding of the underlying causal mechanisms is due, in part, to the fact that most existent studies of these risks have failed to use research designs that can test alternative hypotheses, especially the role of environmental and genetic selection factors. Innovation: We propose to utilize powerful quasi-experimental designs, natural experiments that can pull-apart co-occurring risk mechanisms and account for unmeasured genetic and environmental processes, to explore the effects of several putative early risk factors. In particular, we will compare differentially exposed full siblings, an approach that rules out all genetic factors that could account for the statistical association between early putative risks and later outcomes, as well as all environmental factors that siblings share. We will also compare differentially exposed half siblings, offspring of full siblings (full cousins), offspring of half siblings (half cousins), and offspring of twins. These additional comparisons will enable us to provide converging evidence for our findings and identify the source any familial selection factors (genetic and/or environmental). Detailed assessments of offspring, parental, and community- level risks will further enable us to identify the mechanisms through which early risk factors are associated with severe psychopathology. Approach: We propose to analyze a large dataset that includes information on every individual who was born or lived in Sweden from 1940-1995 (NH7,500,000), which includes measures of early environmental risks, correlated familial and community risks, and well-validated indices of severe criminal, substance use, suicide, and mental health problems from Swedish national registries. Because the dataset also includes information on the biological relatedness of each individual with all of their family members, we are able to conduct numerous quasi-experimental studies of the associations by comparing individuals who differ in both their environmental and genetic risk. Environment and Investigators: Our experienced team includes international experts in the risks associated with early prenatal and perinatal factors; the assessment and etiology of severe psychopathology; the use of large, national registries; and the use of quasi-experimental designs to study putative environmental factors. Indiana University and the Karolinska Institutet provide strong environments and relevant resources.